In 2012, an estimated 22,300 American women will develop ovarian carcinoma (OC) and 15,500 will die of this disease. These statistics highlight the need for improved understanding of the biology of this cancer and improved approaches to therapy. To address this, we have developed treatment-na?ve, intraperitoneally-engrafted, patient-derived xenografts in SCID mice from consecutive patients with ovarian, primary peritoneal, and fallopian tube cancers for the development of novel therapeutics and understanding of OC biology. To date, we have been successful in engrafting over 160 individual models from OC patients of all subtypes, which engraft at a very high rate (~70-75%). These models accurately recapitulate the source patients' tumor histologically and molecularly. Most importantly, the response of Avatar models to cytotoxic chemotherapy is concordant with patient outcomes. Specifically, patients with platinum-resistant OC (PR-OC) have Avatars that do not respond to platinum-based chemotherapy. Conversely, Avatar regress in response to platinum-based chemotherapy originating from patients with platinum-sensitive OC. We now propose to use Avatar models to direct therapy in patients with PR-OC. To reach these goals, we propose to: 1) Development of Avatar models: We will determine the MTD of the four standard salvage agents for patients with PR-OC (topotecan, paclitaxel, gemcitabine, pegylated liposomal doxorubicin). Patients' individual Avatar models will be expanded in the presence of platinum-based chemotherapy, to recapitulate the tumors chemotherapy responsiveness in the patient with PR-OC. To optimize our methodology we will evaluate several interventions aimed at improving the rate of and time to engraftment. In anticipation of accommodating the generation of Avatar models for directing chemotherapy in patients from other institutions, we will assess the feasibility of generating models at a high rate with tumor shipped from the Mayo Clinic-Arizona and Mayo Clinic-Florida to Mayo Clinic- Rochester. 2) Determination of optimal chemotherapy agent for Avatars. We will determine the sensitivity of the individual platinum-resistant Avatar models to the four salvage chemotherapy agents and recommend a 'winning' treatment (or treatments) for each patient at the time she develops PR-OC. Array CGH, SNP and transcriptome profiling will be performed to identify the signature of response to individual agents, which will be enhanced by comparisons among the individual agents to remove generalized chemotherapy responsiveness signature components. 3) Clinical trial of Avatar-directed therapy. Using the individual Avatar response data, treatment will be directed to one of the salvage chemotherapy agents in patients on a phase II clinical trial. Concordance between the Avatar response and patient outcomes will be used to further enrich the signature of response to the four chemotherapy agents. Future studies will then aim to validate the signature.